Storing learnt (no)goods in ROBDDs for solving structured CSPs
نویسندگان
چکیده
It was shown that constraint satisfaction problems (CSPs) with a low width can be solved effectively by structural methods. In particular, the BTD method which exploits the concepts of goods and nogoods makes it possible to solve efficiently difficult instances. However, the memory space required for the storage of these (no)goods may make difficult or impossible the resolution of certain problems. We propose here to represent goods and nogoods with Binary Decision Diagrams (BDD). BDDs are data structures which efficiently represent informations in a compact and canonical form. Then, the practical interest of this trade-off which allows to save space memory to the detriment of time is as-
منابع مشابه
Learning to Identify Global Bottlenecks in Constraint Satisfaction Search
Using information from failures to guide subsequent search is an important technique for solving combinatorial problems in domains such as boolean satisfiability (SAT) and constraint satisfaction problems (CSPs). The information learnt can take various forms such as fine-grained information in the form of no-goods and explanations in CSPs and clause learning in SAT, or coarse-grained informatio...
متن کاملA generic framework for solving CSPs integrating decomposition methods
Many real-world constraint satisfaction problems are structured, i.e., constraints are not uniformly distributed among the set of variables. This structure may be used to improve the solution process of these problems. In particular, backtracking with tree decomposition (BTD) exploits the structure to define variable ordering heuristics and to learn structural goods and nogoods which are used t...
متن کاملA generic bounded backtracking framework for solving CSPs
This paper introduces a new generic backtracking framework for solving CSPs. This scheme exploits semantic and topological properties of the constraint network to produce goods and nogoods. It is based on a set of separators of the constraint graph and several procedures adjustable to exploit heuristics, filtering, backjumping techniques, classical nogood recording, topological (no)good recordi...
متن کاملOptimizing the space to extend the tractability of (valued) structured CSP
It was shown that constraint satisfaction problems (CSPs) with a low width can be solved effectively by structural methods. In particular, the BTD method which exploits the concepts of goods and nogoods makes it possible to solve efficiently difficult instances. However, the memory space required for the storage of these (no)goods may make difficult or impossible the resolution of certain probl...
متن کاملCombining Restarts, Nogoods and Decompositions for Solving CSPs
From a theoretical viewpoint, the (tree-)decomposition methods offer a good approach when the (tree)-width of constraint networks (CSPs) is small. In this case, they have often shown their practical interest. However, sometimes, a bad choice for the root cluster (a tree-decomposition is a tree of clusters) may drastically degrade the performance of the solving. In this paper, we highlight an ex...
متن کامل